Mutations in the Small Gtpase Gene RAB39B Are Responsible for X-Linked Mental Retardation Associated with Autism, Epilepsy, and Macrocephaly

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by AIR Universita degli studi di Milano ARTICLE Mutations in the Small GTPase Gene RAB39B Are Responsible for X-linked Mental Retardation Associated with Autism, Epilepsy, and Macrocephaly

Maila Giannandrea,1 Veronica Bianchi,1 Maria Lidia Mignogna,1 Alessandra Sirri,2 Salvatore Carrabino,2 Errico D’Elia,1,15 Matteo Vecellio,1,16 Silvia Russo,3 Francesca Cogliati,3 Lidia Larizza,3,4 Hans-Hilger Ropers,5 Andreas Tzschach,5 Vera Kalscheuer,5 Barbara Oehl-Jaschkowitz,6 Cindy Skinner,7 Charles E. Schwartz,7 Jozef Gecz,8,9 Hilde Van Esch,10 Martine Raynaud,11 Jamel Chelly,12 Arjan P.M. de Brouwer,13 Daniela Toniolo,2,14 and Patrizia D’Adamo1,*

Human Mental Retardation (MR) is a common and highly heterogeneous pediatric disorder affecting around 3% of the general population; at least 215 X-linked MR (XLMR) conditions have been described, and mutations have been identified in 83 different genes, encoding proteins with a variety of function, such as chromatin remodeling, synaptic function, and intracellular trafficking. The small GTPases of the RAB family, which play an essential role in intracellular vesicular trafficking, have been shown to be involved in MR. We report here the identification of mutations in the small GTPase RAB39B gene in two male patients. One mutation in family X (D-23) introduced a stop codon seven amino acids after the start codon (c.21C > A; p.Y7X). A second mutation, in the MRX72 family, altered the 50 splice site (c.215þ1G > A) and normal splicing. Neither instance produced a protein. Mutations segregate with the disease in the families, and in some family members intellectual disabilities were associated with autism spectrum disorder, epileptic seizures, and macrocephaly. We show that RAB39B, a novel RAB GTPase of unknown function, is a neuronal-specific protein that is localized to the Golgi compartment. Its downregulation leads to an alteration in the number and morphology of neurite growth cones and a significant reduction in presynaptic buttons, suggesting that RAB39B is required for synapse formation and maintenance. Our results demonstrate developmental and functional neuronal alteration as a consequence of downregulation of RAB39B and emphasize the critical role of vesicular trafficking in the development of neurons and human intellectual abilities.

Introduction the role of proteins encoded by these genes led to the hypothesis that XLMR is a synaptopathy-like disorder Human Mental Retardation (MR) is a common and highly that is due, in part, to the fact that most of these proteins heterogeneous pediatric disorder with a severe social localize to pre- and/or post-synaptic neuronal terminals. impact. Accounting for a large, but not well-defined, The identification of guanine nucleotide dissociation portion of all MR forms, genetic defects range from 25%– inhibitor gene, GDI1 (MIM 300104) as one of the genes 50% of the total. Family studies have demonstrated a rela- causing human XLMR suggested that vesicular trafficking tively large number of X-linked forms (XLMR) that seem to is an important pathway for the development of cognitive explain why the incidence of MR in males is about 30% functions.2–4 aGDI, the protein encoded by GDI1, is phys- higher than in females.1 Because large families are not so iologically involved in regulating RAB GTPase proteins. common and many autosomal genes are expected, the More than 60 RAB GTPases are encoded by the human newly identified genes have served as important tools for genome and are localized to different subcellular compart- understanding the molecular basis of MR and have ments.5,6 Through a GDP-GTP exchange cycle, they act as provided novel breakthroughs on the mechanisms and molecular switches to coordinate vesicular transport. The pathways leading to development of the cognitive func- guanine nucleotide-binding status is modulated by three tions altered in MR. classes of regulatory proteins, the guanine nucleotide As a result, at least 215 XLMR conditions and mutations exchange factors (GEFs), GTPase-activating proteins have been characterized in 83 different genes. Research on (GAPs), and the guanine nucleotide dissociation inhibitor

1Dulbecco Telethon Institute at Division of Neuroscience, San Raffaele Scientific Institute, 20132 Milan, Italy; 2Division of Genetics and Cell Biology, San Raffaele Scientific Institute, 20132 Milan, Italy; 3Molecular Genetics Laboratory, Istituto Auxologico Italiano, 20145 Milan, Italy; 4Chirurgia e Odontoiatria Universita` di Milano Polo Osp. San Paolo, 20142 Milan, Italy; 5Max Planck Institute for Molecular Genetics, 14195 Berlin, Germany; 6Practice of Human Genetics, 66424 Homburg (Saar), Germany; 7J.C. Self Research Institute of Human Genetics, Greenwood Genetic Center, Greenwood, SC 29646, USA; 8SA Pathology, Women’s and Children’s Hospital, North Adelaide, SA 5006, Australia; 9Department of Paediatrics, University of Adelaide, 5006 Adelaide, Australia; 10Center for Human Genetics, University Hospital Leuven, 3000 Leuven, Belgium; 11Centre Hospitalier Re´gional Universitaire de Tours, Service de Geńe´tique and INSERM, U930, 37044 Tours, France; 12Universite´ Paris Descartes; Institut Cochin; INSERM, U567; and Centre National de la Recherche Scientifique, UMR 8104, 75014 Paris, France; 13Department of Human Genetics, Nijmegen Centre for Molecular Life Sciences and Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen Medical Centre, 6500 Nijmegen, The Netherlands; 14Institute of Molecular Genetics- CNR, 20182 Pavia, Italy 15Present address: Department of Experimental Oncology, European Institute of Oncology, 20139 Milan, Italy 16Present address: Vascular Biology and Gene Therapy Laboratory, Centro Cardiologico Monzino IRCCS, 20138 Milan, Italy *Correspondence: [email protected] DOI 10.1016/j.ajhg.2010.01.011. ª2010 by The American Society of Human Genetics. All rights reserved.

The American Journal of Human Genetics 86, 185–195, February 12, 2010 185 proteins (GDIs), which retrieved GDP-bound, inactive used for detecting the presence of the protein, and beta-Tubulin RAB proteins from the membranes.7 RAB and RAB-associ- was used for normalizing the total amount of protein. Primer ated proteins have been shown to play an important role sequences are listed in Table S1. in a number of diseases.8 Thus, the association of GDI1 mutations with XLMR led us to hypothesize a role for RAB39B Expression Profile in Human and Mouse X-linked RAB genes in XLMR. Three out of four X-linked Tissues by qRT-PCR RAB genes are specific to the brain, as shown in the Human RNA from various tissues was purchased from Clontech. SymAtlas database, and we identified novel loss-of-func- Mouse tissue RNAs were obtained from different C57BL/6N adult mouse organs (Charles River). Total RNA was isolated with TRIzol tion mutations in the RAB39B (MIM 300774) gene in reagent (Invitrogen) according to the manufacturer’s instructions. two XLMR families. RAB39B is one of the RAB GTPase For the production of cDNA, 1 mg of total RNA was reverse proteins of unknown function, and understanding how transcribed with M-MLV enzyme (Invitrogen) according to the the lack of RAB39B is involved in the pathology of XLMR manufacturer’s instructions. is therefore particularly relevant for understanding the Real-time PCR was performed with SYBR Green Universal Mix pathogenesis of the disease. (Sigma-Aldrich) on a Light Cycler 480 (Roche Diagnostics). Samples were run in triplicate, and data quantification was done

via the comparative threshold cycle (Ct) method with formula Material and Methods 2-(inputCt-controlCt), assuming that the efficiency was 2. Histone H3 with mHistone3F/mHistone3R primers was used for normaliza- Patients and Analysis of the RAB39B Gene tion. Human RAB39B and the mouse Rab39b cDNAs were amplified In total, 22 DNA samples from males with XLMR mapped to Xq28 with the same primer pairs, mR39bF/mR39bR. Mouse Rab39a was were obtained from the European Mental Retardation Consortium amplified with primers R39aF/R39aR. Primer sequences are listed in Nijmegen, The Netherlands, from the Greenwood Genetic in Table S1. Center in North Carolina, USA, and from the MRX72 family in Italy.9 We also analyzed 136 DNA samples from unmapped In Situ Hybridization males with XLMR from the Greenwood Genetic Center and the In situ hybridization was performed on P90 mouse brains with Women’s and Children’s Hospital, Adelaide, Australia and 110 anti-sense and sense RNA probes containing part of the 30UTR of DNA samples from unmapped males with XLMR from the Rab39b, as previously described.3 European Mental Retardation Consortium in Nijmegen, The Netherlands. From these samples, 92 were from patients with Cell Cultures XLMR and autism spectrum disorder, 94 from patients with Primary cultures of hippocampal neurons at embryonic day E18 autism spectrum disorder only, and 94 from Greenwood Genetic were prepared as described by Banker and Cowan.10 Cells were Center patients with epileptic seizures only. grown in MEM supplemented with 10% FBS, 1% glutamine, and Control DNAs were from a previous collection of adult Euro- 1% penicillin and streptomycin. pean males. All patients and controls were recruited and studied after appropriate ethical approval was obtained. Sequences of the two exons of RAB39B were obtained by PCR with the primers cDNA Constructs and MISSION shRNA Validation R39b1exF/R39b1exR and R39b2exF/R39b2exR. The same primers For transfection experiments, different plasmid backbones were were used for direct sequencing on a 3730 automated sequencing generated. apparatus. Analysis was performed with the SeqScape sequence For production of pEGFP-RAB39B, the PCR product obtained analysis software (Applied Biosystems). Sequences were numbered from mouse Rab39b cDNA clone (BC052472) was cloned with according to the RAB39B cDNA NM_171998 and protein primers R39bgfpF/R39bgfpR into XhoI and HindIII sites of NP_741995. Primer sequences are listed in Table S1 available pEGFP-C3 (Clontech). with this article online. For production of pEGFP-RAB39A, the PCR product obtained from C57Bl/6N total brain cDNA was cloned with primers R39AF/R39AR into XhoI and HindIII sites of pEGFP-C3. Mutated cDNA Cloning, Transfection, For production of pFLAG-RAB39B, the PCR product obtained and Immunoblotting from the BC052472 clone was cloned with primers R39bflagF/ Total RNA from a normal subject, family X (D-23), and MRX72 R39bflagR into NotI and KpnI sites of pCMV2-FLAG (Sigma- male patients was extracted from lymphoblastoid cell lines Aldrich). with the RNeasy kit (QIAGEN). RNA was reverse transcribed For generation of pFLAG-RAB39BRescue, the Rab39b sequence of with M-MLV enzyme (Invitrogen). cDNAs were amplified with pFLAG-RAB39B was modified with the QuickChange Site-Directed primers R39b1exF/R39b2exR, inserted into pGEM-T easy vector Mutagenesis Kit (Stratagene) and primer pairs RescueR39bF/Rescu- (Promega), and amplified with specific primers HindhR39bF/ eR39bR. The sequence GCCTACTACAGGAATTCAGTA was re- BamhR39bR so that they could be cloned in frame into HindIII placed with the sequence GCGTATTACCGGAACTCCGTC, which and BamHI sites of pCMV2-FLAG plasmid (Sigma-Aldrich). did not change the amino acid composition of RAB39B but pFLAG-RAB39B constructs were transfected into HeLa cells with rendered the construct insensitive to the Rab39b-specific shRNA. Lipofectamine 2000 (Invitrogen), as recommended by the sup- VSV-pseudotyped third-generation lentiviruses were produced plier. Two days after transfection, cells were harvested and lysed as previously described.11 Rab39b was excised via AgeI and SalI in 1% SDS. Total protein extracts (30 mg) were loaded onto a sites from pEGFP-RAB39B and cloned into the AgeI and SalI 10% polyacrilamide gel and then transferred on a nitrocellulose sites of pCCL.sin.cPPT.PGK.GFP.WPRE plasmid (pCCL-RAB39B) membrane. Polyclonal anti-FLAG antibody (Sigma-Aldrich) was carrying a PGK promoter-GFP cassette.11

186 The American Journal of Human Genetics 86, 185–195, February 12, 2010 For the Rab39b silencing experiments, four MISSION shRNAs TRITC-conjugated phalloidin (Sigma-Aldrich) was used for visual- recognizing different sequences on the Rab39b cDNA (code izing F-Actin. Slides were mounted on Vectashield (VectaLab). TRC000046, 47, 48, and 49) and control scramble shRNA were Images were acquired by UltraVIEW ERS spinning-disk confocal purchased from Sigma-Aldrich. HeLa cells transfected with microscope (Perkin Elmer). pEGFP-RAB39B or pEGFP-RAB39A and transduced with the different shRNAs at a multiplicity of infection (MOI) of 1 or 3 Image Analysis were analyzed by both immunoblot and qRT-PCR so that the ImageJ Analysis Software was used for measuring the area of capacity of shRNAs to specifically knock down Rab39b expression neuronal growth cones. Sholl analysis was performed as previ- could be tested. Three out of four shRNAs were able to downregu- ously described12 so that the degree of neuronal differentiation late Rab39b, but not Rab39a, by about 60%. Control-scramble could be analyzed. shRNA (shScramble), TRC000047, and 49 shRNAs (shRab39b) were introduced into pCCL.sin.cPPT.PGK.GFP.WPRE lentiviral Statistical Analysis vector. Factorial or repeated-measures analysis of variance (ANOVA) with Primer sequences are listed in Table S1. an alpha value of 5% was used (SAS Institute, Cary, NC). Normal distribution and homogeneity of variances among data samples Cell Transfection and Transduction was tested with the K-S normality test and Bartlett’s test for For the intracellular localization of RAB39B by immunofluores- homogeneity. Main effects were verified with t tests. Data points cence experiments, N2A, HeLa, and Cos7 cells were transfected represent the mean, and bars represent the standard error. with Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. So that overexpression of RAB39B did not Animals occur, different amounts of plasmids encoding pEGFP-RAB39B All experiments with animals were performed in strict accordance were tested. The convenient condition was 350 ng for N2A and with the guidelines of the Institutional Animal Care and Use m 100 ng for HeLa and Cos7 cells instead of 4 g as suggested by Committee of the Department of Biological and Technological the supplier. Research, San Raffaele Scientific Institute. So that overexpression of RAB39B in hippocampal neurons did not occur, cells were transduced with pCCL-RAB39B lentivirus at an MOI of 1 immediately after plating. Results For the rescue experiment, N2A cells were transduced with shRab39b or shScramble lentiviral particles at an MOI of 3, and Detection of Mutations in the RAB39B Gene after 4 days they were transfected for 24 hr with 130 ng of We examined 22 XLMR families that mapped to Xq28 to pFLAG-RAB39BRescue or pFLAG-RAB39B. The Rab39b level was determine whether RAB39B was responsible for XLMR. analyzed by qRT-PCR as previously described. From the genomic sequence of RAB39B, primers were de- signed for direct sequencing of the two exons and the BFA Treatment flanking splice sites of this gene. A nonsense mutation Two days after transfection with pEGFP-RAB39B, N2A, HeLa or (NM_171998, c.21C > A) in family X (D-23) introduced Cos7 cells were treated with 2.5 mg/ml of Brefeldin A (Sigma- a premature stop codon seven amino acids after the start Aldrich) for 30 min at 37C. codon (TAA; NP_741995, p.Y7X; Figure 1A). In the MRX72 family, a nucleotide substitution (NM_171998, FM4-64 Uptake Assay c.215þ1G > A) altered the 50 splice junction of intron 1 For FM4-64-uptake experiments, neurons were incubated with (Figure 1B). Using RT-PCR, we demonstrated that a cryptic 10 mM FM4-64 (Molecular Probes) diluted in Krebs-Ringer-HEPES 0 solution (KRH) containing 55 mM KCl (depolarizing solution) 5 splice donor site (GT at 46–47 bp of cDNA, NM_171998) and 1 mM TTX for 1 min at room temperature (RT), rinsed three was activated in the first exon and produced an aberrant times by complete substitution with KRH containing 50 mM mRNA that could lead to synthesis of a putative novel CNQX and 1 mM TTX over a course of 10 min, and then live imaged. peptide of 52 amino acids in length (data not shown). In both cases, immunoblotting of HeLa cells transfected Immunofluorescence and Microscopy with FLAG-tagged mutated and control cDNAs did not Cells were fixed with 120 mM phosphate buffer containing 4% detect any mutant protein (Figure 1C). The mutations paraformaldehyde and 4% sucrose, washed in PBS 1x, and then segregated with the disease in both families as shown in incubated overnight at 4C with primary antibodies diluted in the family pedigrees (Figures 1D and 1E). Sequencing of GSDB 1x (16% Goat Serum, 0.3% Triton X-100, 20 mM phosphate 150 normal chromosomes demonstrated that the muta- buffer, and 450 mM NaCl). Primary antibodies against the tions were not common polymorphisms. following proteins were used for immunofluorescence: GM130 Patients in both families presented with moderate to (BD Transduction), GIANTIN (gift from M.A. De Matteis), severe MR, as well as additional features; in the MRX72 RAB3A, VAMP4, STX16, STX13 (SySy), RAB11 (Transduction Labo- family, III-9, III4, and III-5 had seizures, and IV-5 presented ratories), Flag (Sigma), AP2 (Affinity BioReagents, clone AP6), 9 SYN1 (gift from F. Valtorta), and SHANK1 (gift from C. Sala). After with autism spectrum disorder (Figure 1E). In family X cells were washed in a high-salt solution (20 mM phosphate buffer (D-23), four brothers (III-1, III-2, III-3, and III-4) and two and 500 mM NaCl), appropriate rhodamine (TRITC)- or Cy5- maternal uncles (II-3 and II-4) were mentally retarded, conjugated secondary antibodies (Jackson Laboratories) were and these six patients presented with macrocephaly, as added for 1 hr at RT. DAPI was used for visualizing cell nuclei. described in Table 1.

The American Journal of Human Genetics 86, 185–195, February 12, 2010 187 seizures and autism spectrum disorder. We found the same synonymous variation (c.822A > G, p.T181T) that Tarpey et al.13 described previously in two unrelated autistic males.

RAB39B Is Highly Enriched in the Brain We performed qRT-PCR on several human and mouse tissues to analyze RAB39B expression. We demonstrated that RAB39B is enriched in the brain (Figures 2A and 2B). During postnatal (P) mouse brain development, Rab39b expression is constant between P1 and P10, as well as between P20 and P180. A significant difference was observed between P10 to P20, where Rab39b was upregu- lated 2-fold as calculated by the t test (p ¼ 0.003). We also analyzed the expression profile of the Rab39a gene, which has 72% and 76% identity at the cDNA and protein levels, respectively. The Rab39a transcript was expressed at levels ten times less than Rab39b and does not increase at all stages analyzed. These results demonstrated that Rab39b is a brain-specific gene with an important role (Fig- ure 2C). Expression analysis of the Gdi1 null brains revealed no change, suggesting that absence of aGDI did not alter Rab39b RNA levels, as previously shown for other RAB proteins3 (data not shown).

Rab39b Is a Neuron-Specific Gene Rab39b appeared to be ubiquitously expressed in brain, as determined by in situ hybridization on sagittal sections of adult (P90) mouse brain. The highest expression was observed in the hippocampus, as shown in Figures 3A–3F. We further analyzed the Rab39b neuronal cellular specificity. We dissected and dissociated P2 mouse cerebral cortices to obtain different neuronal cell types such as microglia, oligodendrocytes, astrocytes, neuronal precursors, and neurons, as described.14 By qRT-PCR, we quantified the Rab39b expression in these cell types and showed that Rab39b was specifically expressed in neuronal precursors and neurons only (Figure 3G).

Figure 1. RAB39B Mutations in Family X (D-23) and MRX72 (A and B) The upper panels show the corresponding sequence RAB39B Localizes to the Golgi Compartment from a normal individual; the lower panels are from the patients. We analyzed the intracellular localization of RAB39B in Arrows indicate the mutations. Sequence numbers refer to the pEGFP-RAB39B-transfected Cos7 and HeLa cells (data not human RAB39B cDNA (NM_171998). shown), and in a pCCL-RAB39B-transduced neuroblas- (C) Immunoblot analysis performed on 30 mg of total protein extracted from HeLa cells transfected with pFLAG-RAB39B con- toma cell line, N2A (Figure S1a), and primary hippocampal structs containing the wild-type RAB39B cDNA (WT), family X neurons (Figure 4). In all instances, RAB39B appeared (D-23) mutated cDNA, or MRX72 cDNA. The 24-kDa band corre- to localize to the Golgi compartment, as judged from the sponds to FLAG-RAB39B fusion protein, and the 51-kDa band colocalization with the Golgi markers, GIANTIN and corresponds to beta-tubulin. MOCK: cells transfected with pCMV2-FLAG. NT: cells not transfected. GM130. We conﬁrmed the Golgi localization by treating (D) Pedigree of family X (D-23). N2A cells for 30 min with or without brefeldin A (þ BFA (E) Pedigree of MRX72. Black squares indicate males with mental and BFA), respectively (Figure S1b). GFP-RAB39B and retardation, dotted circles indicate carrier females, and open anti-GIANTIN immunostaining, as well as GFP-RAB39B symbols indicate unaffected individuals. Asterisks indicate family members who have been sequenced in this study. immunostaining after BFA treatment, showed dispersion of the Golgi compartment. We also observed a partial Because of the presence of seizures and autism spectrum colocalization with markers that cycle from the cell sur- disorder in family X (D-23) and MRX72, we analyzed 526 face to the trans-Golgi network (TGN) via sorting and DNA samples from unmapped X-linked patients with recycling endosomes such as VAMP4, SYNTAXIN16, and

188 The American Journal of Human Genetics 86, 185–195, February 12, 2010 Table 1. Clinical Evaluation of Family X (D-23)

Clinical Feature II-3 II-4 III-1 III-2 III-3 III-4

Age (yr) 52 50 19 16 13 13

Height (cm) 163 170 178.5 177.5 145.5 145.7 (<3rd percentile) (3rd percentile) (50th percentile) (50th percentile) (3rd percentile) (3rd percentile)

OFC 59 62.5 60.5 56.5 57 56.5 (>97th percentile) (>97th percentile) (>97th percentile) (>97th percentile) (>97th percentile) (>97th percentile)

Degree of MR mild severe moderate moderate moderate severe

Additional features - - obesity - autism autism

In family X (D-23), four brothers and two maternal uncles were mentally retarded. The pedigree of the family is shown in Figure 1D. Patients III-3 and III-4 were twins. A sister of the two uncles (II-7) was mentally normal and died of a brain tumor at the age of 21 years. The degree of mental retardation ranged from mild (II-3) to severe (II-4 and III-4). All six patients had macrocephaly. Brain MRI scans in patients III-3 and III-4 and a CT scan in patient III-1 revealed no abnormalities. Height ranged from short stature in II-3, II-4, III-3, and III-4 to normal height in III-1 and III-2. Clinical examination of the patients revealed neither additional malformations nor signiﬁcant facial dysmorphic features.

RAB11.15–17 The latter subcellular colocalization suggests Downregulation of Rab39b Gene Alters Neuronal a possible role of RAB39B in the recycling pathway con- Differentiation necting the plasma membrane, endosomes, and TGN. To mimic the loss-of-function mutations identified in patients, we knocked down Rab39b expression in mouse Downregulation of Rab39b by shRNA primary hippocampal neurons. Immediately after plating, Undifferentiated N2A cells were infected with two inde- hippocampal neurons were transduced with shRab39b and pendent shRNAs against Rab39b at an MOI of 3. As deter- shScramble lentiviral particles at an MOI of 1. ShRAB39b mined by qRT-PCR after 5 days of infection, both shRNAs was able to reduce Rab39b expression by 40%, as deter- were able to downregulate the endogenous Rab39b expres- mined at 3 days in vitro (DIV) by qRT-PCR. Morphological sion by 45%, expressed as n-fold changes (5 standard error analysis from 80 shScramble and 124 shRab39b randomly [SE]) relative to untreated cells (shRab39b: 0.55 5 0.02). In sampled images from four independent experiments contrast, shScramble did not have any effect (shScramble: showed a significant alteration in morphology and the 0,91 5 0.006) (Figure S2a). number of growth cones (GCs; Figure 5A). GCs were The specificity of shRab39b was determined by a knock- decreased in number at each neurite terminal. The mean down-rescue experiment. N2A cells were transduced with number of GCs per neurite (5SE) was 3.1 (50.1) for the shRab39b lentiviral particles for 4 days and then trans- shRAB39b and 3.8 (50.1) for shScramble neurons. The fected with the full-length wild-type Rab39b (pFLAG- difference was significant (ANOVA: F [1,202] ¼ 8.2; RAB39B) or pFLAG-RAB39BRescue constructs. In cells p ¼ 0.005). The mean area of GCs (5SE) was 38.7 (52.8) treated with shRab39b and transfected with pFLAG- for shRab39b and 30.8 (52.4) for shScramble neurons. RAB39B, the wild-type pFLAG-RAB39B expression was The difference was significant (ANOVA: F [1,202] ¼ 3.8, downregulated by 45% in comparison to pFLAG- p ¼ 0.05) (Figures 5B and 5C; see the enlargement in the RAB39B-transfected cells as determined by qRT-PCR right panel). (shRab39b þ pFLAG-RAB39B: 0.55 5 0.1; value expressed Sholl analysis,12 as a measure of neuronal differentiation, as n-fold changes 5 SE). We also showed that shRab39b was performed on 62 shScramble and 87 shRab39b was not able to downregulate the rescue Rab39b transcript randomly sampled images from three independent experi- (Figures S2b and S2c). ments and also revealed a significant decrease in the After 4 days of treatment with the shRab39b lentivirus, number of neuronal branches: repeated-measures ANOVA we observed a change in N2A cell morphology. To further between treatments had a value of F [1,147] ¼ 6.6, investigate the morphological phenotype, we stained N2A p ¼ 0.01 and showed an interaction between treatment shRab39b lentivirus-treated cells with phalloidin as a and Sholl category (mm from the cell body): F [1,5] ¼ 2.3, marker for actin cytoskeleton morphology. We found p ¼ 0.04. Factorial ANOVA for each Sholl category revealed clumps of actin and reduced and/or altered membrane significant p values for 30 mm (F [1,147] ¼ 9.3, p ¼ 0.003), protrusions that were not present in shScramble-treated 35 mm (F [1,147] ¼ 7.6, p ¼ 0.006), and 40 mm (F [1,147] ¼ N2A cells (Figure S2d). After 4 days, transfection of 7.2; p ¼ 0.007) (Figure 5D). Taken together, these data shRab39b-treated N2A cells with pFLAG-RAB39BRescue suggest a global alteration in neurite extension. for 24 hr was able to restore the normal actin cytoskeleton morphology. RAB39B Was Localized to GCs These results suggested that the downregulation of To test whether RAB39B was present in GCs, we transduced Rab39b might alter intracellular vesicular trafficking steps primary hippocampal neurons at 3 DIV with pCCL- important for membrane remodeling. RAB39B lentiviral particles. We found that GFP-RAB39B

The American Journal of Human Genetics 86, 185–195, February 12, 2010 189 Figure 3. Rab39b Is Specifically Expressed in Neurons Figure 2. Expression Analysis on Different Human and Mouse (A) In situ hybridization analysis of Rab39b on P90 mouse brain. Tissues Light-microscopic images showing the distribution of Rab39b m (A) Expression of RAB39B transcript in different human tissues. mRNA in 20 m mouse brain sagittal sections. Anti-sense (A) (B) Expression of Rab39b transcript in mouse tissues. and sense (B) Rab39b probe. (C) Expression profile of Rab39b and Rab39a in mouse total brain (C–F) Enlargements of specific regions: CC, cerebral cortex; CbC, during post-natal (P) development (from P1 to P180; for each age cerebellar cortex; Hi, hippocampus; OB, olfactory bulb; T, thal- point, n ¼ 8). Data were expressed as Rab39b or Rab39a expression amus; S, subiculum; CA1 and CA3, fields of the hippocampus; normalized to Histone-H3 (2-DCt(Rab39b–H3))(5SE). DG, dentate gyrus; PL, Purkinje layer; and Mi, mitral cells. Scale bars represent 2 mm (A and B) and 0.1 mm (C–F). (G) Rab39b expression normalized to Histone-H3 (2-DCt(Rab39b–H3)) signals were spread in the whole GC area, as shown by in neuronal cell types. P2 mouse cerebral cortices were dissected, phalloidin staining (Figure 6A), and some discrete signals dissociated in trypsin medium, and plated. Different neuronal cell types were prepared as described.14 Rab39b is specifically colocalized with the Golgi-derived vesicles, as shown in expressed in neuronal precursors and neurons. the double labeling with GIANTIN (Figure 6B). All these data suggested that RAB39B could be involved in intracel- gulation of endogenous Rab39b RNA levels was quantified lular trafficking related to membrane recycling of Golgi- at 7 DIV by qRT-PCR, and we found that the Rab39b derived vesicles; its reduction may impede such trafficking expression was 70% reduced in shRab39b-treated neurons and thus lead to an accumulation in the membrane compared with shScramble-treated neurons. content and disorganized growth of GCs. Analysis of SYNAPSIN1-positive puncta, a marker of the presynaptic compartment, revealed 49% reduction in RAB39B Is Involved in Synapse Formation the mean number of synapses. The quantification of the Next, we looked at the subsequent formation of pre- mean number of synapses (5SE) obtained from 56 synaptic terminals of primary neurons transduced shRab39b-treated neurons was 115 (59.2), compared immediately after plating with shRab39b and shScramble with 223 (517.7) from 53 shScramble randomly sampled lentiviral particles at an MOI of 1 until 7 DIV. The downre- images from three independent experiments. Factorial

190 The American Journal of Human Genetics 86, 185–195, February 12, 2010 ANOVA revealed F [1,97] ¼ 32.7 and p < 0.0001 (Figures 7A and 7B). We then measured FM4-64 uptake to analyze the ability of the synaptic vesicles in the remaining synapses to engage in exo- and endocytosis. The quantiﬁcation of 85 shRab39b and 93 shScramble randomly sampled images from three independent experiments revealed that the recycling of synaptic vesicles appeared to be unaffected because no gross differences in exo- or endocytosis activity were observed. Instead, FM4-64-positive labeling (5SE) showed a 40% reduction in shRab39b neurons (163 5 8.9) compared to shScramble-treated (272 5 12) ones, and a signiﬁcant difference was observed by factorial ANOVA (F [1,151] ¼ 50.7, p < 0.0001) (Figure 7C). These data suggest that the alteration observed in neurite extension at 3 DIV might lead to defective synapse formation or instability.

Discussion

The four X-linked RAB genes were considered as candidate genes for XLMR. Mutation analysis was started on the RAB39B gene, and here we have demonstrated that when mutated, the gene can cause XLMR. We report the identification of two independent point mutations in two unrelated families affected with XLMR. These mutations disrupt the synthesis of RAB39B by introducing a premature stop codon in the open reading frame. The mutations cosegregated with the disease and were not found in 150 controls. Some of the affected males from both families primarily presented with mental retardation associated with autism spectrum disorder, epileptic seizures, and macrocephaly. This adds further evidence that XLMR is a highly heterogeneous condition in which the genetic background and/or environmental factors contribute to determining the clinical phenotype. More than 60 RAB GTPases are encoded by the human genome and localize to different subcellular compartments.6 Through a GDP-GTP exchange cycle, they act as molecular switches to coordinate vesicular transport. The association of these proteins and their regulators with many diseases reflects the physiological importance of RAB GTPases, which is not surprising in light of their central role in membrane trafficking. To date, mutations in two RAB GTPases have been found in association with hereditary neurological disorders. RAB7 (MIM 602298) Figure 4. RAB39B Intracellular Localization in Mouse Hippo- mutations are responsible for Charcot-Marie-Tooth type campal Neurons 2B neuropathy (CMT2B [MIM 600882])18 and ulcero-muti- Neurons were transduced at 3 DIV with pCCL-RAB39B lentiviral lating neuropathy,19 and Rab23 (MIM 606144) mutations particles (green) at an MOI of 1 and analyzed at 7 DIV. The major colocalization was with GM130 (cis-Golgi) and GIANTIN are responsible for Carpenter syndrome (ACPSII [MIM 20 (Golgi complex), VAMP4 and SYNTAXIN16 (STX16)—involved 201000]) characterized by cranial-suture defects. Among in trafficking from the cell surface to the trans-Golgi network X-linked genes and in addition to GDI1 (MIM 300104), (TGN)—and SYNTAXIN13 (STX13) and RAB11, which are present mutation analysis in the three X-linked RAB genes could in recycling endosomes. No colocalization was observed with the adaptor protein 2 (AP2). The scale bar represents 10 mm. be helpful to highlight new XLMR genes and further evaluate the prevalence of mutations in this gene family in MR. Along the same lines, we should also mention the finding of mutations in the Rab3 p130 GAP gene

The American Journal of Human Genetics 86, 185–195, February 12, 2010 191 (MIM 602536) in Warburg Micro Syndrome (WARBM [MIM 600118])21 and Martsolf Syndrome (MIM 212720),22 severe autosomal-recessive disorders characterized by developmental defects and mental retardation that are possibly also due to an alteration in neuronal intracellular trafficking. The exact cellular localization, tissue expression profile, and physiological function of many RAB GTPases is still unknown, and this is the case for RAB39B. We showed that Rab39b is a brain-enriched gene that is specifically expressed in neurons. A polyclonal antibody against the C terminus hypervariable amino acid sequence of RAB39B was produced but failed to work in immunofluorescence experiments. We investigated the cellular localization by using a GFP-RAB39B fusion protein, and we tried to minimize the overexpression by using lentiviral infection at an MOI of 1 and at least three different cell types. We observed a diffuse staining in the cytoplasm and in the nucleus in our experiments; this staining was possibly caused by GFP-RAB-fused proteins, as shown by several papers reporting a similar diffuse pattern after transfection of RAB proteins to determine intracellular localization.23,24 We cannot demonstrate that this is the case, but we think this is an unspecific staining possibly due to GFP tagging. We showed that GFP-RAB39B fusion protein partially colocalized with known proteins of the Golgi compartment and with markers that cycle from the cell surface to the trans-Golgi network (TGN) via sorting and recycling endosomes. The latter subcellular colocalization suggests a possible role for RAB39B in the recycling pathway connecting the plasma membrane, endosomes, and TGN. Searching for interacting RAB39B proteins led to the identification of different proteins, including one Golgi-associated protein (unpublished data). These preliminary data further support the localization of RAB39B to the Golgi compartment. We investigated the effect of loss-of-function mutations Figure 5. RAB39B Is Involved in Neurite Differentiation identified in XLMR patients. We used the RNA interference Hippocampal neurons were transduced with lentiviral particles containing shRab39b and shScramble at an MOI of 1. The down- technology (shRNA) to knock down the Rab39b gene in regulation of the endogenous Rab39b RNA level was quantified at mouse primary hippocampal neurons and investigate a 3 DIV by qRT-PCR, and 40% reduction was observed in shRab39b- possible effect on neuronal development and connectivity. transduced neurons. To test a possible off-target silencing effect25 and downre- (A) There was a significant difference (p ¼ 0.005) in the mean number of growth cones (GCs) at each neurite terminal (5SE) gulation specificity, we used three shRNAs targeting in shRab39b neurons (3.1 5 0.1) versus shScramble neurons different portions of the Rab39b cDNA in N2A cells and (3.8 5 0.1) (mean 5 SEM). performed a knockdown-rescue experiment. Our data (B) There was a significant difference (p ¼ 0.05) in the mean area showed that all shRNAs were able to downregulate the of GCs (5SE) in shRab39b neurons versus shScramble neurons (38.7 5 2.8 [mean 5 SEM] and 30.8 5 2.4, respectively). expression of Rab39b but not Rab39a, which shares high (C) TRITC-Phalloidin immunostaining revealed altered GC sequence homology with Rab39b. N2A shRNA-treated cells morphology (enlargement of GC morphology in right panel). revealed clumps of actin and altered membrane protru- The scale bar represents 10 mm. sions that returned to normal in the knockdown-rescue (D) For Sholl analysis,12 as a measure of neuronal differentiation, a mask of concentric circles 5 mm apart was created and superim- experiment, as a model of membrane remodeling. posed on neuronal cells. The first 10 mm from the center of the mask corresponds to the neuronal cell body. Repeated-measures ANOVA revealed significant differences between treatments, value for 30 mm [F (1,147) ¼ 9.3, p ¼ 0.003], 35 mm [F (1,147) ¼ where F (1,147) ¼ 6.6, p ¼ 0.01, and between treatment and Sholl 7.6, p ¼ 0.006], and 40 mm [F (1,147) ¼ 7.2; p ¼ 0.007]. category (mm from the cell body), where F (1,5) ¼ 2.3, p ¼ 0.04. *p < 0.05; **p < 0.01; ***p < 0.001. Data are expressed as Factorial ANOVA for each Sholl category revealed a significant p means 5SE.

192 The American Journal of Human Genetics 86, 185–195, February 12, 2010 Figure 6. RAB39B Localizes to Golgi- Derived Vesicles (A) Hippocampal neurons transduced immediately after being plated with pCCL-RAB39B lentiviral particles at an MOI of 1 and immunostained at 3 DIV. TRITC-Phalloidin and anti-Tuj1 immunostaining revealed that RAB39B occupied a region similar to that of the GC actin domain. (B) A partial colocalization was observed with GIANTIN, as a marker for the Golgi- derived vesicles, as shown by the arrows in the enlargement in the merged right panel. The scale bar represents 10 mm.

ficking and lead to an accumulation in the membrane and disorganized growth of GCs. The observed alteration in neurite extension might be the cause of defective synapse forma- In primary hippocampal neurons, Rab39b downregulation or instability, as shown in adult hippocampal tion resulted in the alteration of growth cone morphology neurons. Rab39b downregulation resulted in a drastic and neuronal branching and a global alteration in neurite reduction in the total number of synapses without gross arborization. Localization of RAB39B to the growth cones, differences in synaptic-vesicle recycling in the remaining and the essential role of Golgi-derived vesicles for the synapses. successful generation of a functional GC,26 suggest that In conclusion, we identified mutations in the neuron- RAB39B could be involved in intracellular trafficking specific RAB39B gene, a new gene for XLMR associated related to membrane recycling throughout Golgi-derived with autism spectrum disorder, epileptic seizures, and vesicles. The reduction of RAB39B might impede such traf- macrocephaly.

Figure 7. RAB39B Is Involved in Synapse Formation Hippocampal neurons were transduced immediately after being plating with shRab39b and shScramble lentiviral particles at an MOI of 1. The downregulation of the endogenous Rab39b RNA level was quantiﬁed at 7 DIV by qRT-PCR, and 70% reduction was observed in shRab39b- transduced neurons. (A and B) SYNAPSIN1 immunostaining and quantiﬁcation showed a 49% reduction in the SYN1 presynaptic compartment on shRab39b compared to shScramble neurons. (C) FM4-64-positive labeling showed a 40% reduction for shRab39b neurons compared to shScramble. The scale bar represents 10 mm. ***p < 0.001. Data are expressed as means 5SE.

The American Journal of Human Genetics 86, 185–195, February 12, 2010 193 We showed that RAB39B seems to play a key role in Knevett, S.G., Ramsay, M.F., et al. (2002). Deletion of the synapse formation, possibly by mediating an intracellular mental retardation gene Gdi1 impairs associative memory pathway connecting the plasma membrane, endosomes, and alters social behavior in mice. Hum. Mol. Genet. 11, and TGN and possibly in the organization of neuronal 2567–2580. circuits. Further experiments are needed to elucidate the 4. Bianchi, V., Farisello, P., Baldelli, P., Meskenaite, V., Milanese, M., Vecellio, M., Muhlemann, S., Lipp, H.P., Bonanno, G., biological function of RAB39B and the alterations in Benfenati, F., et al. (2009). Cognitive impairment in Gdi1-defi- neuronal organization and connectivity caused by its cient mice is associated with altered synaptic vesicle pools and absence. short-term synaptic plasticity, and can be corrected by appropriate learning training. Hum. Mol. Genet. 18, 105–117. Supplemental Data 5. Pereira-Leal, J.B., and Seabra, M.C. (2000). The mammalian Rab family of small GTPases: Definition of family and subfamily Supplemental Data include two figures and one table and can be sequence motifs suggests a mechanism for functional speci- found with this article online at http://www.cell.com/AJHG. ficity in the Ras superfamily. J. Mol. Biol. 301, 1077–1087. 6. Pereira-Leal, J.B., and Seabra, M.C. (2001). Evolution of the Rab family of small GTP-binding proteins. J. Mol. Biol. 313, Acknowledgments 889–901. We thank the subjects who participated in this study and their 7. Pfeffer, S., and Aivazian, D. (2004). Targeting Rab GTPases to parents. We also thank L. Wrabetz and B. Oostra for their helpful distinct membrane compartments. Nat. Rev. Mol. Cell Biol. discussion, comments, and critical reading of the manuscript; 5, 886–896. R. Cassinari for graphic support; ALEMBIC for the microscopy 8. Corbeel, L., and Freson, K. (2008). Rab proteins and Rab- technical support; L. Naldini, L. Sergi Sergi for the help in lentiviral associated proteins: Major actors in the mechanism of production; and A. De Matteis for providing us with a surplus of protein-trafficking disorders. Eur. J. Pediatr. 167, 723–729. antibodies. This work was supported by funding from Telethon, 9. Russo, S., Cogliati, F., Cavalleri, F., Cassitto, M.G., Giglioli, R., Italy (TCP04015 to P.D.). M.G. is supported by a fellowship from Toniolo, D., Casari, G., and Larizza, L. (2000). Mapping to the Italian Ministry of Research and Health (PRIN 2007). V.B. is distal Xq28 of nonspecific X-linked mental retardation supported by a Ph.D. fellowship from Vita Salute San Raffaele MRX72: Linkage analysis and clinical findings in a three- University, Milan, Italy. This study was also supported, in part, generation Sardinian family. Am. J. Med. Genet. 94, 376–382. by grants from the National Institute of Child Health and 10. Banker, G.A., and Cowan, W.M. (1977). Rat hippocampal Human Development (HD26202 to C.E.S.) and the South Carolina neurons in dispersed cell culture. Brain Res. 126, 397–425. Department of Disabilities and Special Needs. 11. De Palma, M., and Naldini, L. (2002). Transduction of a gene expression cassette using advanced generation lentiviral Received: October 25, 2009 vectors. Methods Enzymol. 346, 514–529. Revised: December 28, 2009 12. Sholl, D.A. (1950). A discussion on the measurement of Accepted: January 11, 2010 growth and form; The theory of differential growth analysis. Published online: February 11, 2010 Proc. R. Soc. Lond. B Biol. Sci. 137, 470–474. 13. Tarpey, P.S., Smith, R., Pleasance, E., Whibley, A., Edkins, S., Hardy, C., O’Meara, S., Latimer, C., Dicks, E., Menzies, A., Web Resources et al. (2009). A systematic, large-scale resequencing screen of The URLs for data presented herein are as follows: X-chromosome coding exons in mental retardation. Nat. Genet. 41, 535–543. X-linked mental retardation, http://www.ggc.org/xlmr.htm and 14. Butovsky, O., Ziv, Y., Schwartz, A., Landa, G., Talpalar, A.E., http://xlmr.interfree.it/home.htm Pluchino, S., Martino, G., and Schwartz, M. (2006). Microglia Gene expression, http://symatlas.gnf.org/SymAtlas/symform activated by IL-4 or IFN-gamma differentially induce neuro- Online Mendelian Inheritance in Man (OMIM), http://www.ncbi. genesis and oligodendrogenesis from adult stem/progenitor nlm.nih.org/Omim cells. Mol. Cell. Neurosci. 31, 149–160. Statistic analysis, http://www.statview.com 15. Khvotchev, M.V., Ren, M., Takamori, S., Jahn, R., and Sudhof, European Mental Retardation Consortium, http://www.euromrx. T.C. (2003). Divergent functions of neuronal Rab11b in Ca2þ- com regulated versus constitutive exocytosis. J. Neurosci. 23, 10531–10539. 16. Amessou, M., Fradagrada, A., Falguieres, T., Lord, J.M., Smith, D.C., Roberts, L.M., Lamaze, C., and Johannes, L. (2007). Syn- References taxin 16 and syntaxin 5 are required for efficient retrograde 1. Herbst, D.S. (1980). Nonspecific X-linked mental retardation I: transport of several exogenous and endogenous cargo A review with information from 24 new families. Am. J. Med. proteins. J. Cell Sci. 120, 1457–1468. Genet. 7, 443–460. 17. Tran, T.H., Zeng, Q., and Hong, W. (2007). VAMP4 cycles from 2. D’Adamo, P., Menegon, A., Lo Nigro, C., Grasso, M., Gulisano, the cell surface to the trans-Golgi network via sorting and M., Tamanini, F., Bienvenu, T., Gedeon, A.K., Oostra, B., Wu, recycling endosomes. J. Cell Sci. 120, 1028–1041. S.K., etal. (1998). Mutationsin GDI1 are responsible for X-linked 18. Verhoeven, K., De Jonghe, P., Coen, K., Verpoorten, N., non-specific mental retardation. Nat. Genet. 19, 134–139. Auer-Grumbach, M., Kwon, J.M., FitzPatrick, D., Schmedding, 3. D’Adamo, P., Welzl, H., Papadimitriou, S., Raffaele di Barletta, E., De Vriendt, E., Jacobs, A., et al. (2003). Mutations in M., Tiveron, C., Tatangelo, L., Pozzi, L., Chapman, P.F., the small GTP-ase late endosomal protein RAB7 cause

194 The American Journal of Human Genetics 86, 185–195, February 12, 2010 Charcot-Marie-Tooth type 2B neuropathy. Am. J. Hum. Genet. D.T., Stoodley, N., et al. (2006). Mutation in Rab3 GTPase-acti- 72, 722–727. vating protein (RAB3GAP) noncatalytic subunit in a kindred 19. Houlden, H., King, R.H., Muddle, J.R., Warner, T.T., Reilly, with Martsolf syndrome. Am. J. Hum. Genet. 78, 702–707. M.M., Orrell, R.W., and Ginsberg, L. (2004). A novel RAB7 23. Barbero, P., Bittova, L., and Pfeffer, S.R. (2002). Visualization mutation associated with ulcero-mutilating neuropathy. of Rab9-mediated vesicle transport from endosomes to the Ann. Neurol. 56, 586–590. trans-Golgi in living cells. J. Cell Biol. 156, 511–518. 20. Jenkins, D., Seelow, D., Jehee, F.S., Perlyn, C.A., Alonso, L.G., 24. Itoh, T., Fujita, N., Kanno, E., Yamamoto, A., Yoshimori, T., Bueno, D.F., Donnai, D., Josifova, D., Mathijssen, I.M., and Fukuda, M. (2008). Golgi-resident small GTPase Rab33B Morton, J.E., et al. (2007). RAB23 mutations in Carpenter interacts with Atg16L and modulates autophagosome forma- syndrome imply an unexpected role for hedgehog signaling tion. Mol. Biol. Cell 19, 2916–2925. in cranial-suture development and obesity. Am. J. Hum. 25. Rao, D.D., Vorhies, J.S., Senzer, N., and Nemunaitis, J. (2009). Genet. 80, 1162–1170. siRNA vs. shRNA: Similarities and differences. Adv. Drug Deliv. 21. Aligianis, I.A., Johnson, C.A., Gissen, P., Chen, D., Hampshire, Rev. 61, 746–759. D., Hoffmann, K., Maina, E.N., Morgan, N.V., Tee, L., Morton, 26. Erez, H., Malkinson, G., Prager-Khoutorsky, M., De Zeeuw, J., et al. (2005). Mutations of the catalytic subunit of RAB3GAP C.I., Hoogenraad, C.C., and Spira, M.E. (2007). Formation of cause Warburg Micro syndrome. Nat. Genet. 37, 221–223. microtubule-based traps controls the sorting and concentra- 22. Aligianis, I.A., Morgan, N.V., Mione, M., Johnson, C.A., tion of vesicles to restricted sites of regenerating neurons after Rosser, E., Hennekam, R.C., Adams, G., Trembath, R.C., Pilz, axotomy. J. Cell Biol. 176, 497–507.

The American Journal of Human Genetics 86, 185–195, February 12, 2010 195